What role does artificial intelligence play in content recommendations?


PHOTO: Adobe


Marketers see great potential value in using artificial intelligence (AI) to support the use case of recommending highly targeted content to users in real time. This use case scored the highest among the 49 use cases presented to marketers in the State of AI Marketing 2021 Report by Drift and the Marketing Artificial Intelligence Institute.

This use case scored 3.96, putting it on the “high value” point (4.0), with 5.0 being “transformer”. Top five ranked AI marketing use cases include:

  • Adapt audience targeting based on look-alike behavior and analysis (3.92)
  • Measure ROI by channel, campaign and globally (3.91)
  • Discover insights into top performing content and campaigns (3.86)
  • Create Data Driven Content (3.82)

“Most of the websites you visit today for business, a human writes the rules for what content to recommend,” Paul Roetzer, CEO and Founder of Institute of Artificial Intelligence Marketing, told CMSWire in a CX decoded podcast. “What are the related articles? There is a basic tagging system for if they read this, then read that. Most of them are human powered. They don’t have a Netflix or Spotify-like algorithm that actually learns preferences, knows the last 15 articles someone read, and how far they entered. It does not pull any other kind of behavioral intention data. Most don’t.

The data conundrum

That’s where the potential lies, but it’s something marketers and customer experience professionals are hopeful of: 54% of them told CMSWire researchers in the State of the Digital Customer Experience Report 2021 they see AI having significant impacts on the digital customer experience over the next two to five years. And most of them see “getting actionable customer insights” (27%) as the area in which they see the most potential.

Roetzer said it was difficult to come up with very good solutions to do this immediately. Noz Urbina from Urbina Consulting agreed, calling the nascent technology.

The bigger question for marketers beyond the type of tools available is whether we have the data to support the use case, according to Roetzer. And do we have a solid foundation of metadata, content markup, and content taxonomies, according to Urbina.

“You need enough data, for one,” Roetzer said. “Sometimes the problem is with the smaller data, not necessarily the cost. Do you have enough data to make it worth trying to use a machine learning algorithm to do it better than a human? Do you have enough traffic on your site to justify it? “

Associated article: CX Decoded Podcast: Practical Use Case of AI in Marketing

Build or buy?

Does it make more sense to build a bespoke solution on AWS or Google, or is there an out-of-the-box solution to plug in for a few thousand dollars a month that will teach our users and start making recommendations? These are some questions marketers should ask themselves when considering using AI for targeted content recommendations, according to Roetzer.

“A lot of people actually build on GPT-3, a technology that came out of OpenAI, which was kind of a lab developed to quickly advance AI technology and then share it with the world, so Open AI in the name, ”Roetzer said.

According to OpenAI, nine months after the launch of the first commercial product, the OpenAI API offers more than 300 applications. These 300 or so companies are developing language generation capabilities on the backbone of GPT-3, according to Roetzer. He quoted conversion.ai and copy.ai, the latter who made sure $ 2.9 million in funding in March. “What they (copy.ai) do is they have a bunch of pre-trained models, so you just get a subscription, and you can actually go in there, give it some input … and he’ll write in made a copy of the ad for you, an email copy. Very interesting.”

OpenAI officials cited the example of Algolia, which has partnered with OpenAI to integrate GPT-3 with its advanced search technology to create their new product “Answers” that better understands customer questions and concerns. connects to the specific part of the content that answers their questions, according to OpenAI officials.

“Algolia Answers helps publishers and customer support services query in natural language and bring up non-trivial responses,” they wrote. “After performing GPT-3 tests on 2.1 million news articles, Algolia found an accuracy of 91% or better and Algolia was able to accurately answer complex questions in natural language. “

Associated article: 8 considerations when selecting an AI marketing provider

Respond to behaviors

Urbina said the most popular method of generating targeted content in real time via AI is through recommendation engines. According to Google developers, content-based filtering uses item features to recommend other items similar to what the user likes, based on their previous actions or explicit comments.

“Rather than just responding to a query, which is a definition of a search engine, recommendation engines respond to behaviors,” Urbina said. “Your location, your activities, your previous search behaviors, all of these things are ambient data that search engine technology can use and then turn into a recommendation engine. And artificial intelligence and machine learning, of course, are the basis of that. They will find patterns in the data and then make the appropriate recommendations. “

According to Urbina, the most common phrase after the recommendation engine that marketers need to know is “the next best action.” Marketers want to walk people through the journey, and machine learning helps determine various subsequent actions, such as sending an email, texting, or offering recommended content that is displayed.

“And he’s optimizing that on a scale beyond human capabilities,” Urbina said. “So the AI ​​needs to observe user activity and, based on that user’s data correlated with what all the other users are saying, determine the best thing I could suggest to move them through the process. travel, so that’s basically the main area we need to focus on for content recommendations: finding out how we can correlate user behaviors against trends to establish the next best action, which can be recommended content. . ”

AI can’t do it alone

According to Urbina, marketers often struggle to let AI do the heavy lifting. There is a need to have a solid structured content plan in place: tag and apply metadata to existing content, then create content taxonomies.

“One of the most successful things we’re doing now is developing the taxonomy so that the recommendation engine has something to work with,” Urbina said. “A taxonomy of personas. A taxonomy of business scenarios. A taxonomy of challenges. A taxonomy of benefits. A taxonomy of functions. A taxonomy of content types. A taxonomy of channels. place this taxonomy that defines these compartments, what can AI work with? ”

With a foundational content structuring program, marketers can define what makes a white paper, what makes a case study, what makes a brochure, what makes a product preview, etc. Urbina.

“That said, if you don’t have any taxonomy, you can unleash AI on whatever content there is,” Urbina said. “It can read it, use natural language processing to see what the topics and commonalities the words are in your content, and you can then organize that to become your real taxonomy. So before you start you can use AI to determine what your potential taxonomies are and what your categorizations might be. Either way, a human needs to guide the AI. “

Why a “brute force” approach to AI won’t work

Most marketers just want a data scientist to do all of these things, but that isn’t always possible. Plus, they often don’t really realize that if they’ve been involved in organizing the content around taxonomies, metadata, and markup, the whole operation will be much more efficient.

“And that’s where I see this technology is absolutely nascent. And its effectiveness through brute force approaches is what slows it down the most, ”Urbina said.

Can this AI technology supporting content recommendation AI engines take the brand tone? Roetzer said he’s getting there.

“He’s made giant strides over the past three years,” Roetzer said. “2013 was kind of that tipping point where AI caught up with the promise of what it could possibly do, and language is at the heart of that. This is why voice assistants have become good. This is why some chatbots have gotten good, and why understanding and language generation has reached entirely new levels in recent years. And so the ability to understand and reproduce the tone, if it’s not there, it comes. And there are a lot of people who put a lot of money behind this kind of thing. “

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *